One hundred and five of 395 patients with acute pancreatitis were surgically treated in our clinic from 1981 to 1984. Ninety three of these patients were examined with contrast enhanced computed tomography and/or ultrasound and were clinically assessed according to Ranson's objective criteria before operation. At operation, 77 patients showed necrotising pancreatitis and 16 showed biliary acute interstitial pancreatitis. Ninety per cent of the cases with extensive and 79% of those with minor necroses of the pancreas had been demonstrated with contrast enhanced computed tomography. Ultrasound failed to be diagnostic in 24% of the patients due to meteorism; the sensitivity of the diagnostic studies for pancreatic necrosis was 73% regardless of the extent of the process. Using the early objective signs, seven patients with acute interstitial pancreatitis were classified as having a severe attack, whereas 30 patients with necrotising pancreatitis were categorised as mild attacks. We conclude that the contrast enhanced computed tomography is an aid in deciding on conservative or surgical treatment in a case of acute pancreatitis. Ultrasound does not appear to be an adequate method for determining pancreatic necrosis. The early objective signs fail to sufficiently identify the necrotising form of acute pancreatitis.